Immune metabolic changes identify causal candidate genes and enable diagnostic frameworks in MAFLD.
Journal:
Scientific reports
Published Date:
Aug 28, 2025
Abstract
Metabolic dysfunction-associated fatty liver disease (MAFLD), a global epidemic affecting 25% of adults, is driven by immune-metabolic dysregulation, yet the causal mechanisms linking immune cell-specific gene perturbations to disease progression remain unresolved. Current studies lack systematic integration of single-cell transcriptomics, causal inference, and functional validation to dissect actionable potential intervention targets. We combined peripheral blood mononuclear cells (PBMCs) single-cell RNA sequencing (scRNA-seq; GSE179886: 2 MAFLD vs. 4 controls) with two-sample Mendelian randomization (MR; GWAS data: 8,434 cases vs. 770,180 controls) to prioritize causal candidate genes. Machine learning (101 algorithms) and multi-cohort validations (GSE126848, GSE63067, GSE89632) established diagnostic models. Causal candidate gene expression and functional impact were validated in high-fat diet (HFD)-fed mice, ob/ob mice, AML12 hepatocytes, and primary hepatocytes. scRNA-seq identified 212 differentially expressed genes (DEGs) across six immune cell types, with CD4 + T cells and monocytes showing the most significant dysregulation (FDR < 0.001). MR analysis revealed 37 causal candidate genes, including PRF1 (protective: IVW OR = 0.68, 95% CI 0.59-0.79; p = 1.2 × 10⁻⁵) and EVI2B (risk: OR = 1.42, 95% CI 1.21-1.67; p = 3.8 × 10⁻⁴), which antagonistically modulated MAFLD risk. A machine learning model integrating five causal candidate genes (PRF1, EVI2B, CST7, GNG2, KLHL24) achieved robust diagnostic accuracy (training AUC = 1.00; validation AUC = 0.74-0.78), outperforming conventional biomarkers. In vivo validation in both HFD-fed and ob/ob mice confirmed marked overexpression of PRF1, EVI2B, CST7, GNG2, and KLHL24 in hepatic tissue (p < 0.05), with EVI2B overexpression significantly exacerbating lipid accumulation in AML12 and primary hepatocytes. This study pioneers the integration of scRNA-seq, MR, and cross-species and cellular validation to unravel immune-driven metabolic dysfunction in MAFLD. We identify EVI2B as a pro-steatotic driver and provide a causally informed diagnostic framework grounded in experimental validation, advancing mechanistic understanding toward future targeted interventions.
